productive and unproductive labor in the us...

Productive and Unproductive Labor in the US Economy: Does the Distinction Matter?

A reply to Houston and Laibman

Simon Mohun

Department of Economics, Queen Mary and Westfield College,

Mile End Road, London E1 4NS.

e-mail: [email protected] telephone: +44-207-882-5089

November 1999

Abstract: Responding to criticisms of Mohun (1996) by Houston (1997) and Laibman (1999), this paper

explores the consequences of the competing definitions proposed for productive labor, and

concludes that abandoning the productive-unproductive labor distinction renders the Marxian

theory of capitalist accumulation an unconvincing account of US accumulation from 1948 to

1989.

JEL Classification: E11, E32, N12

Keywords: Productive labor, unproductive labor, rate of profit, US economy

Simon Mohun Productive and Unproductive Labor in the US Economy 18 December, 2015

2

Reflections on a theoretical debate

I would like to thank the two Davids, Houston (1997) and Laibman (1999) for their comments on my paper Mohun

(1996). Their criticisms are similar in the main, questioning whether it makes any sense to distinguish ‘production’

from ‘circulation’ and to confine the production of value and surplus value to the former. I will not repeat the

arguments of these earlier papers.1 Instead, the purpose of this reply is to try to show why the productive/unproductive

labor distinction matters.

I want to begin with the questions with which Laibman (1999) finishes. He asks,

“Why is labor the source and substance of value? What validity criteria should be applied to the substantiation

of this proposition? What role does this proposition play in the theory of the nature and dynamics of capitalist

society?” (Laibman 1999, p.72)

and he talks of “the urgent need to pose these basic questions” (ibid.) The reason why labor has a special place in

Marxian theory is uncontroversial, being a ‘primitive postulate’ of historical materialism: purposive productive labor

is the essence of what it means to be human; this activity alters the environment within which such labor occurs, and

this alteration in turn alters the producers themselves. 2 In and of itself, this does not demonstrate that labor is the

source and substance of value, for there can be no such demonstration other than assertion. But it does predispose one

to think in those terms. All that can be done with this understanding at a theoretical level is to assert (or to deny) that

such assertions are reasonable. Whether they are acceptable (the “validity criteria”) depends on what can be done with

them. And one test (perhaps the most important test) of the success of such a construction is empirical; that is, can a

theory be built upon the foundation of the labor theory of value that generates propositions which enable us to

understand the historical evolution of capitalism as it has actually occurred? And this is simultaneously to approach

1 Save, that is, for corrections. Laibman comments (Laibman 1999, p.64, n.4) that I am in two minds as to how to classify

non-waged labor. The definition (in Mohun 1996 p.36, Proposition 4) that non-wage labor is neither productive nor

unproductive, but non-productive, was intended to refer to housework and similar activities, and I should have called it

non-remunerated labor. Non-wage labor in Table 1 (ibid. p.40) is explicitly labor which receives remuneration which is

different from wages (either the labor of employers or that of the self-employed).

2 Of course theorists who deny the existence of ‘essences’ will find this a controversial statement.


3

Laibman’s third question, for unless it can be specified how the labor theory of value can be used to understand ‘the

nature and dynamics of capitalist society’, it should be abandoned.

Hence it is my view that the ‘urgent need’ is not to pose questions about theoretical postulates (which are probably

unanswerable), but rather to investigate the historical development of capitalism from a Marxist perspective. If we

have a theory that says that its fundamental variables might rise, or fall, through time, then it is not obvious how to

confront the propositions of that theory with empirical evidence. The theory might organise the interpretation of the

evidence, but the evidence cannot be used to say that the theory is ‘right’ or ‘wrong’. Putting the point more strongly,

if theory and reality are different, so that empirical evidence is an issue, then theory is different from a metaphysical

belief system. In the latter, there are limitless immunising strategies available to insulate beliefs from confrontation

with reality, whereas with a theory it is important to be able to specify, at least in principle, what sort of evidence

would lead one to reject that theory. Consider then the Marxist theory of capitalist development. It postulates general

trends, and while these trends can be reversed, such reverses are countervailing tendencies, which do not dominate the

general trend. In particular, the theory postulates the following inter-related (tendential) propositions:

• the rate of surplus value will rise as absolute and relative surplus value are produced;

• as relative surplus value is produced, then, for a given money wage, both the value of money and the value of

labor-power will fall;

• with the productivity increases which produce relative surplus value, real wages will tend to rise;

• productivity increases are typically the outcome of ‘labor-saving capital using’ technical change, via

concentration and centralisation; if technical change is of this type, then it must generate increases in the

technical and organic compositions of capital;

• productivity increases will produce some negative feedback via price reductions, but these will not be so

great as to produce a fall in the value composition of capital; i.e. the increase in quantities through

accumulation will tend to dominate the price reduction effects of productivity increases;

• because of the bias of technical change, the reserve army of labor will tend to rise, along with relative

immiserisation;


4

• in the course of establishing these secular tendential developments, accumulation will tend to proceed

cyclically, and the rate of profit will tend to fall (at least cyclically).

Marxist discussion tends to focus on the theoretical logic of the postulated bias of technical change and its effects on

the composition of capital, and on the analytical causes of a falling rate of profit. These are not my concern here.

Rather, if the above propositions form a ‘picture’ of capitalist development, then it is important to investigate the

extent to which there is empirical support for that picture.3

That is why the approach in Mohun (1996) is explicit in its advocacy of the approach suggested in this journal by

Foley (1982).4 Indeed, that 1996 paper was written in the course of an attempt at an empirical analysis of the

Australian economy,5 and in the construction of the paper I therefore saw propositions 1-6 as interlinked in that

endeavor. Laibman (1999) is only concerned with propositions 3 and 4; he is not concerned with the empirical

literature, but rather with the coherence of the theory itself, arguing that the empirical literature is meaningless if the

theoretical distinction between productive and unproductive labor is neither coherent nor valid. Houston (1997) uses

the complexity of reality to suggest that the abstractions used in Mohun (1996) are inappropriate.

Theoretical coherence is obviously important. But I have little further to say about the theoretical issues. The criteria

for whether labor is productive or unproductive are straightforward. Does an activity produce a new use-value or alter

an existing one in some way? Does that process of production produce surplus value, or does it consume it? For those

who assert that all wage labor is productive, then all such labor does produce a new use-value, or does alter an existing

one, and it does produce surplus value. Hence all ‘circulation’ activity is productive of a use-value that is a bearer of

3 While one might doubt the logic behind the derivation of some of the postulates, they might nevertheless obtain as

empirical phenomena.

4 Laibman (1999, p.71) remarks that the categories of productive and unproductive labor do not appear in Foley (1982). But

he could consult Foley’s 1986 book, ch.7, especially pp.116-24.

5 See Mohun (1998).


5

value and surplus value, and there is no process employing wage labor that does not produce surplus value.6

Conversely, for those who assert that some wage labor is unproductive, then that labor does not produce a new

use-value, or does not alter an existing one, and it does not produce surplus value, but rather consumes it. In particular,

all ‘circulation’ activity is by definition unproductive, transferring commodity capital into money capital and thence

into productive capital. These are definitions, and the “precise and convincing deduction” (rather than “an exercise in

semantic persuasion”) which Laibman requires (Laibman 1999, p.69, n.8) is just not possible.

For example, if privately employed wage labor in circulation activities is productive of value and surplus value, then

the argument of the first six chapters of Capital I is difficult. To make sense of it, one has to argue either that the

subject matter is simple commodity production (Houston 1997) or that it is a heuristic designed purely to get at the

process of production (Laibman 1999). Houston’s approach forces one into a controversial logico-historical

methodology; Laibman’s approach is more arbitrary, because it would make equal sense as a heuristic to assume

equality of the values of inputs and outputs in production and to focus on the increase in value in circulation. These

approaches of Houston and Laibman, then, no more and no less than that of Mohun (1996), are part of a ‘vision’ of the

process of capitalism. If that vision is to convince, then it must show that it has something sensible, indeed interesting,

to say about actual capitalist development. For this is the only game in town, and without informed empirical analysis,

Marxian theory is of no interest. If Marxism is to regain any credibility, it is going to have to show that it can say

something empirical and meaningful about modern capitalism and its historical evolution.

The remainder of this paper is empirical. It attempts to show that a Marxist approach that uses the

productive-unproductive labor distinction is consistent with the basic vision suggested by Marx’s theory of capitalist

accumulation (at least over the greatest period of accumulation in the history of capitalism to date), and that it enables

one to think fruitfully about the changing structure of an economy through time. Conversely, one that ignores the

productive-unproductive labor distinction is inconsistent with that theory, at least in its application in the second half

6 Laibman (1999, p.63, n.3) ignores the problem of workers in the state sector, and gives no hint as to how he sees them in

this context. More generally, one might question whether a category such as the rate of exploitation is well-defined under

circumstances in which the production of value continues to the point of consumption; and one might question Laibman’s

explication of the category ‘use-value’ in his remark “does an object to which you do not have title have the same

use-value as one to which you do?” (ibid., p.62)


6

of the twentieth century. This poses a challenge to Marxists who would abandon productive-unproductive labor

distinction, for it suggests that combining such an abandonment with adherence to the Marxian vision of accumulation

is not possible (unless that vision is reduced to the uninteresting proposition that variables might rise or fall through

time).

US capitalism 1948-89

Since the categories ‘productive/unproductive’ are defined relative to the production of value and surplus value, any

empirical analysis has first to propose some account of how the value magnitudes of Marxian theory are related to

observable empirical data. Several approaches can be found in the literature. One is simply to assume some

correspondence (Weisskopf 1979). A second is to reorder the flows recorded in national accounts and input-output

tables to produce monetary magnitudes of Marxian variables, and to use regression analysis to show an approximate

correspondence between labor values and prices (Shaikh and Tonak 1994, Cockshott et al. 1995). A third is to use

theoretical considerations to assert an exact correspondence between labor value aggregates and money value

aggregates (Foley 1982, and rather differently, Moseley 1991, and contributors to Carchedi and Freeman 1996).

Whichever is adopted will structure the interpretation of the empirical evidence. In Mohun (1994, 1996), the

accounting framework proposed by Foley (1982) is followed, and this will be continued below. This enables me to say

that the profit-wage ratio is identical to the rate of surplus value, and to interpret the fixed capital stock to wages ratio

as (a proxy for) the value composition of capital.7

Looking at macroeconomic time series, one obvious variable on which to focus is the rate of profit (r). Any formal

macroeconomic model of the economy has to generate explanations/predictions that are compatible with observed

time series. If they are to be other than accounting frameworks, such models must propose behavioural hypotheses,

but for present purposes it is sufficient to concentrate on the accounting framework itself, and this requires in turn

some meaningful decomposition of the rate of profit. First, define the rate of profit as the ratio of profits (Π ) to the

private sector capital stock (K), where profits are the difference between (adjusted) Net Domestic Product (NDP) and

wages, and the capital stock is defined in net rather than gross terms. Two adjustments are made to NDP. First,

imputed rent is subtracted (because it does not correspond to any flow of value), and second, the wages, salaries and

7 A proxy, since the circulating capital elements of constant capital are omitted.


7

supplements of state sector workers are subtracted (because they produce no marketed output).8 Call such adjusted

NDP ‘money value added’ or MVA. Then the rate of profit is defined as

KWMVA

Kr −

==Π (1)

This is illustrated below for the US economy from 1948 to 1989.

(Chart 1 about here)

Broadly, the rate of profit falls from 1948 to 1958, rises from 1958 to 1965, falls from 1965 to 1982, and rises from

1982 to 1989, reaching its mid-1970s level by the end of the 1980s. In considering decompositions of the rate of

profit, I will consider two cases, one in which all labor is productive, and one in which some labor is unproductive. In

both cases, I will assume that all labor is employed in capitalist firms under capitalist conditions.9

i) All labor is productive

If all labor is productive, an obvious decomposition of the rate of profit is to divide numerator and denominator by

total wages, interpreting the numerator as the rate of surplus value and the denominator as the value composition of

capital. That is,

WK

WK

rΠΠ

== (2)

8 To say that state sector workers produce no marketed output is a stylization, for it is increasingly the case (at least in the

UK) that some (though not full) charges are made for state services where this is possible. Nevertheless, the stylization

follows standard national accounting practice (based on SNA 93), which defines the general government sector as

institutional units engaged in the non-market production of output for collective and individual consumption, financed out

of compulsory payments, and/or engaged in redistribution.

9 This is not an adequate treatment either of the self-employed or of capitalists themselves, but a ‘broad brush’ treatment

will suffice for present purposes.


8

Movements in the rate of profit can be accounted for by the differences in the movement of these two ratios. This is

attractive because there is a sustained theoretical treatment of the movement of these ratios in Capital that can be used

as a starting point for the translation of an accounting framework into a theory.10

This decomposition is illustrated in Table 1, showing these decompositions for the US economy from 1948 to 1989, as

both a cumulative change over the whole period, and as changes within two profit cycles 1948-65 and 1965-89.11

[Table 1 about here]

The two profit cycles are rather different. In the earlier one, the dominating effect in the downswing is the rise in the

composition of capital K/W, and in the upswing is the rise in the rate of surplus value Π/W, but over the whole cycle

the rise in K/W dominates. In the later and rather longer cycle, movements in Π/W dominate, and to such an extent that

they continue to dominate when both cycles are considered together. That is, of the overall 50 per cent fall in the rate

of profit, three-fifths is accounted for by the shift from profits to wages, and the remaining two-fifths by the rise in the

composition of capital.

Table 2 shows the same decomposition, but over NBER dated cycles.12


10 Other decompositions are obviously possible. For example, one could divide numerator and denominator by MVA, so that

movements in the rate of profit are accounted for by the sum of movements in the aggregate profits share (used as an

indicator of class struggle) and in average capital productivity (the inverse of the capital-output ratio, used as an indicator

of the capital intensity of production). I will restrict attention to the decomposition of equation (2), purely because of its

‘classical’ provenance in the history of Marxism.

11 Peaks are denoted by P and troughs by T.

12 The NBER dates the last peak in July 1990. Since the data set I use ends in 1989, I have used 1989 as the final peak. Note

also that I have used 1979 as a peak rather than the NBER date of January 1980. It is interesting to note that prior to 1970,

turning points in the series for the rate of profit are not at all well correlated with NBER turning points; from 1970

onwards, they are so.


9

With the exception of the 1969-73 cycle when there was a negligible total rise of a quarter of one per cent, the rate of

surplus value fell in every cycle up to and including the 1980 trough. Thereafter it has risen. And with the exception of

a small fall in the 1957-60 cycle, the composition of capital rose across every cycle until the 1981 peak, after which it

fell sharply through the 1980s.13 The two ratios are illustrated in Chart 2.

(Chart 2 about here)

This is not a picture that is easy to reconcile with standard Marxist theory. The competitive struggle to increase

productivity through innovation is supposed to drive up the composition of capital, and this is supposed to increase

relative surplus value and hence the rate of surplus value. Hence there is a presupposition that the composition of

capital and the rate of surplus value will be positively related. But the US history is one of a generally inverse relation

between the rate of surplus value and the composition of capital (the only exceptions being the 1957-60 and 1969-73

cycles). Table 3 illustrates with the variance-covariance matrix.


Assuming that volatility is measured by variance, much the greater component in the volatility of the annual rate of

change in the rate of profit is the variance of the annual rate of change in the rate of surplus value. In terms of

associations between the series, the correlation matrix tells the same story.


Annual rates of change in the rate of profit are closely correlated with annual rates of change in the rate of surplus

value, and are negatively correlated with annual rates of change in the composition of capital, but the correlation with

the latter is only about two thirds that with the former (in absolute terms). Movements in the rate of surplus value

dominate those of the composition of capital in their effects on the rate of profit. Moreover, whereas the theory

predicts a positive correlation between the annual rate of change of the composition of capital and the annual rate of

change of the rate of surplus value, the empirical evidence shows this correlation to be negative.

13 A similar picture emerges from the decomposition into profit share and capital productivity: consistently falling profit

share (apart from a small rise over the 1969-73 cycle) and falling capital productivity (apart from a small rise over the

1957-60 cycle) until the Reagan years, which then see a rising profit share and rising capital productivity.


10

The way in which Marxists have traditionally dealt with such prima facie incompatibility of theory and evidence is

either to ignore the evidence, or to assert that the evidence is ‘really’ different from what it appears to be, or to add

something (some countervailing condition perhaps) which can render the evidence compatible with the theory. The

first route of ignoring evidence is to substitute (usually dogmatic) metaphysics for science, and is of little interest. As

regards the second route, that the evidence is ‘really’ different from what it appears to be, Laibman (1993) has rightly

criticised attempts to construct Marxian empirical measures which have no linkage to “real processes” (Laibman 1999

p.70).14 Hence all that is left is to add something.

Consequently, to retain the Marxist account of both long run secular and shorter run cyclical capitalist development,

the obvious addition is to assert that, as a countervailing tendency, movements in the rate of profit reflect the changing

balance of class forces. During periods when the rate of profit falls (1948-58 and 1965-82), the US working class must

have been sufficiently economically strong to wrest all productivity gains, and more, from the capitalist class. And

when the rate of profit rises, (1958-65 and 1982-89) the reverse must have been the case, but over the whole period,

the former predominates. Given the light regulation of the US labor market and its weak labor market institutions, one

possible indicator of working class strength is the tightness of the labor market: the lower the rate of unemployment

the more is the working class able to appropriate productivity gains, and hence the lower is real profits per hour. The

correlation of real profits per hour against the unemployment rate is 0.54, so that there is indeed a positive association;

but the scatter diagram of Chart 3 does not show a close enough association for this story to be entirely convincing.

[Chart 3 about here]

Moreover, the correlation of the aggregate profit-wage ratio against the unemployment rate lends no support to this

story, because while the association is quite strong (an absolute value of 0.73), the sign is negative, which is just the

opposite of what would be required. Finally, a more general association with a periodization of world capitalism is not

strong either, for inspection of Chart 1 shows that the US rate of profit fell for a further ten years after the end of the

‘golden age’. And the correlation between aggregate profit-wage ratio and the unemployment rate for the period 1948

to 1973 is negative and weaker at -0.39.

14 Perhaps I was not sufficiently clear, but Houston (1997) has misunderstood my position. The only rate of profit is that

defined by equation (1). There is no ‘Marxian’ rate of profit that is different, more basic, underlying, or whatever.


11

To summarise, if all labor is assumed productive, the empirical evidence from the USA does not appear to be

consonant with the basic propositions of the Marxian theory of accumulation. A story of working class strength could

be added, as a determining countervailing factor, but a convincing story of this has yet to be told, for the evidence is

not strong. The addition of working class strength as a countervailing factor is therefore vulnerable to the charge of

constructing an immunising strategy in order to avoid a potentially fatal conflict between the Marxian theory of

accumulation and the empirical evidence. One might assert that

“Marx’s C-M-C vs. M-C-M′ distinction is a brilliant insight into two forms of market relations, and his

questions surrounding the origin of surplus value were a pathbreaking means of investigating the connection

between the surface structure of formal equality in exchange and a deep structure of exploitation and

coercion.” (Laibman 1999, p.68)

Yet the analytics of the Marxian theory of accumulation which build on this insight and these questions (capital as

‘self-expanding value’) do not appear to be robust as an account of capitalist development when confronted with

empirical evidence which is constructed on the basis that all labor is productive.

ii) Only some labor is productive

Now assume that labor is either productive or unproductive, as asserted in Mohun (1996). Call the wages paid to

productive labor W(p) and those to unproductive labor W(u). Then the rate of profit can be written as

)(

)()(

)()}({

)()(

pWK

pWuW

pWpWMVA

KuWpWMVA

Kr

−−

=

−−==

Π

(3)

The first term in the numerator is the rate of surplus value, and the second is the ratio in wage terms of unproductive to

productive labor; the denominator is (a proxy for) the value composition of capital.15 Movements in the rate of profit

15 This is not quite right, because unproductive capital is included in the numerator of the composition of capital. However,

commercial and financial capital has little fixed capital stock relative to productive capital, and so it is unlikely there will

be too much distortion.


12

are then accounted for by movements in the numerator and denominator of equation (3). Notice that the definitions of

the rate of surplus value and the composition of capital in equation (3) are different from those in equation (2). Table 5

describes US post-war profit cycles in terms of this decomposition.


As with the previous story, within each cycle, the composition of capital rises in the downswing and falls in the

upswing, but, unlike in the previous story, over each cycle as a whole, and over both cycles taken together,

movements in the composition of capital dominate. Over each cycle as a whole and over the whole period, the

composition of capital and the rate of surplus value move in the same direction, although the movement of the rate of

surplus value in the second cycle’s downswing remains anomalous. Table 6 shows the same decomposition, but over

NBER dated cycles.


Once labor employed in circulation activities is considered unproductive, then over the whole period the rate of profit

falls by some 49 per cent, accounted for by a rise in the composition of capital by some 53 per cent, and a rise in the

rate of surplus value by 40 per cent which is offset by a rise of 70 per cent in the wage ratio of unproductive to

productive labor. The only cycle in which the rate of surplus value does not rise is 1969-73 cycle, which both sees the

end of the ‘golden age’ in the international context and is perhaps the most turbulent period of US post-war history.

With the exception of the 1969-73 and the 1981-89 cycles, the composition of capital rises and the rate of surplus

value rises across each cycle; and the only significant falls in the rate of surplus value are in the upswing of 1970-73

and the downswing of 1975-79. And the substantial rise in the rate of profit during the 1980s is accounted for by a

long sustained rise in the rate of surplus value and a large increase in capital productivity as the composition of capital

falls steadily. All of this is compatible with standard Marxist theory. There is little evidence in the data for any

strength of the US working class vis-à-vis capital. Rather, the wage ratio of unproductive to productive labor, which

rises in every phase except the upswings of 1970-73 and 1975-79, provides information on the changing structure of

the US economy, as increasing amounts of surplus value are required to convert commodity capital into money capital

and thence into productive capital.


13

Turning now to volatility, Table 7 shows that the variance in the annual rate of growth of the composition of capital,

when not all labor is productive, is nearly double that when all labor is assumed productive. Annual rates of change in

the rate of surplus value and the wage ratio of unproductive to productive labor are considerably lower, although their

difference (the numerator of equation (3)) is high.


In terms of associations between the series, the correlation matrix in Table 8 tells a similar story.


In annual growth rate terms, the numerator of equation (3) is less positively correlated with the rate of profit than the

numerator of equation (2), and the denominator more negatively correlated. As one would expect, the annual rate of

growth of the rate of profit is positively correlated with that of the rate of surplus value, and negatively correlated with

that of the wage ratio of unproductive to productive labor, and with that of the composition of capital. Moreover, the

growth in the rate of surplus value and the growth in the composition of capital are both positively associated with the

growth in the wage ratio of unproductive to productive labor (correlation coefficient of 0.46 and 0.75 respectively).

This suggests a story, at least until the early 1980s, of productivity growth driven by labor-saving capital-using

technological change, combined with growing difficulties in the sphere of circulation; labor is displaced from

employment in the production of value and simultaneously is sucked into employment in the sphere of circulation.

Such labor is financed by the rise in productivity in the productive sector, leaving little net gain for capital overall.

Hence the rise in the composition of capital drives the rate of profit down.

By contrast, following the trough of 1982, the negative effects of strong growth in the wage ratio of unproductive to

productive labor are swamped by the rise in capital productivity and the associated rise in the rate of surplus value, a

restructuring of US capital after some thirty years of decline, which produces a large rise in the rate of profit.

Finally, in Mohun (1996, p.51, equation (12)) I proposed a decomposition of what I have called above ‘the rate of

surplus value when all wage labor is productive’, calling it the ‘class distributive’ rate of exploitation e(d), which I

reproduce here:


14

−

+=

)()(

)()()()(

pWuW

VS

pWuWpWde (4)

The three constituent ratios and e(d) are illustrated in Chart 4.

[Chart 4 about here]

Contrary to Laibman’s scepticism (Laibman 1999, p.70), I maintain my assertion that this does indeed “provide a

starting point for a rich agenda of empirical investigation” (Mohun 1996, p.51), for it explicitly directs attention to the

interaction between the rate of surplus value on the one hand, and structural change on the other.

Conclusion

This paper has had the aim of investigating the extent to which the historical record of some forty years of US

capitalism provides broad evidence for the Marxian theory of capital accumulation. Implicit in the paper is the

argument that decomposing the rate of profit into constituent elements and investigating movements of these latter is a

legitimate accounting exercise. The paper contrasts two such decompositions, one in which all private sector wage

labor is productive, and the other in which only some such labor is productive. Of course, the wage ratio of

unproductive to productive labor in equation (3) is not a ratio to which capitalists relate; but then neither is the profit

share of national income, nor the composition of capital, nor the profit-wage ratio. The point is to use a decomposition

that is informative. Critics of the productive/unproductive labor distinction argue that it is arbitrary and hence not

informative. But the empirical evidence argues for the opposite. The paper concludes that abandoning the distinction

between productive and unproductive labor, and defining all private sector wage labor as productive, at the same time

as maintaining the broad outlines of the Marxian theory of capitalism, is not possible, unless some implausible

immunising stratagem is proposed. If one thinks that Marxian theory provides a vision that is in some sense true, and

one includes in that vision definitions of productive and unproductive labor, then the empirical measurement of such

definitions yields results which are broadly consonant with what the Marxian theory of capitalist development would

lead one to expect, at least for the longest and most sustained period of accumulation in the history of capitalism. This

is not in itself evidence that the theory is true. But it is suggestive that a Marxian theory built on ignoring the

distinction between productive and unproductive labor is not very useful.


15

Yet the contrast sketched in this paper is too crude. It is over-schematic, indeed ‘undialectical’, to contrast a

worldview in which all labor is productive and ‘class struggle’ determines everything, with one in which not all labor

is productive and ‘capital’ determines everything. For in the former, the tendential laws of capitalist development

disappear, and in the latter, class struggle is reduced to an epiphenomenon of the self-expansion of capital. A theory in

the Marxian tradition which is empirically informative must clearly embrace both. In pursuit of this, the paper

suggests that one interesting way in which the changing structure of the US economy can be examined using Marxian

categories is through the wage ratio of unproductive to productive labor (or equivalently through the wage ratio of

productive to total labor). This opens up an area of conjunctural analysis that has been little explored. Within a

different but related framework, Moseley (1991, ch.5) has suggested a focus on commercial labor and its relatively

slower productivity growth compared with that of productive labor, which required a rising proportion of the labor

force to sell the more rapidly growing output of productive labor. And Duménil and Lévy (1999) have more generally

proposed a periodization that sees the emergence of a new ‘managerial’ mode of production, as management is

separated from property ownership, and finance and credit relationships assume a greater predominance. But more

remains to be done, both for the US and for other developed capitalist economies, both to investigate and to develop

the claims of Marxian theory as a guide to the analysis of developed capitalism.

Appendix: Data Sources

The data is annual, and is all taken from secondary sources, which suffices for the purposes of this paper. The major

difficulty in the primary sources lies in the calculation of the wage bills of productive and unproductive labor. I have

relied on the calculations presented by Shaikh and Tonak (1994), which are sufficiently similar to what I would need

to calculate along the lines of Mohun (1996). But I have excluded Wages to Government Workers from the series for

Wages to Unproductive Workers. While there is some element of arbitrariness empirically in making the

productive/unproductive distinction, I do not believe that marginally different choices would alter significantly the

argument of the text.

Series for GNP, NNP, GDP, Imputed Rent, and Wages to Government Workers, to Productive Workers and to

Unproductive Workers are all taken from the appendices to Shaikh and Tonak (1994). The Capital Stock series is from


16

Duménil and Lévy (1993). The dating of NBER cycles is taken from Romer (1999). The unemployment rate is from

BLS. MVA is computed as GDP less capital consumption (taken as the difference between GNP and NNP) less

Imputed Rent less Wages to Government Workers. Annual and total rates of growth are computed as differences in

natural logs divided by the time span over which the growth is taken. All calculations were performed using Microsoft

Excel in Office 97.

References

Cockshott, Paul, Allin Cottrell and Greg Michaelson. 1995. “Testing Marx: Some New Results from UK Data”.

Capital & Class 55(Spring): 103-134.

Duménil, Gérard. 1983-84. “Beyond the Transformation Riddle: A Labor Theory of Value”. Science & Society

XLVII(4): 427-450.

Duménil, Gérard and Dominique Lévy. 1993. The Economics of the Profit Rate. Aldershot UK: Edward Elgar.

Duménil, Gérard and Dominique Lévy. 1999. “Periodizing Capitalism. Technology, Institutions and Relations of

Production”. Mimeo.

Foley, Duncan K. 1982. “The Value of Money, the Value of Labor Power, and the Marxian Transformation Problem”.

Review of Radical Political Economics 14(2): 37-47.

Foley, Duncan K. 1986. Understanding Capital. Cambridge, Mass: Harvard University Press.

Freeman, Alan and Guglielmo Carchedi (eds.). 1996. Marx and Non-Equilibrium Economics. Cheltenham UK:

Edward Elgar.

Houston, David. 1997. “Productive-Unproductive Labor: Rest in Peace”. Review of Radical Political Economics

29(1): 131-39.

Laibman, David. 1993. “The Falling Rate of Profit: A New Empirical Study”. Science and Society 57(2): 223-233.

Laibman, David. 1999. “Productive and Unproductive Labor: A Comment”. Review of Radical Political Economics

31(2): 61-73.

Lipietz, Alain. 1982. “The So-Called ‘Transformation Problem’ Revisited”. Journal of Economic Theory 26(1): 59-88.


17

Mohun, Simon. 1994. “A Re(in)statement of the Labor Theory of Value”. Cambridge Journal of Economics 18:

391-412.

Mohun, Simon. 1996. “Productive and Unproductive Labor in the Labor Theory of Value”. Review of Radical

Political Economics 28(4): 30-54.

Mohun, Simon. 1998. “Unproductive Labor and the Rate of Profit in Australia 1966/67-1991/92”. Chapter 16, in

Bellofiore, R. (ed.), Marxian Economics: A Reappraisal, Vol. II. Basingstoke: Macmillan, 1998,

pp. 252-69.

Moseley, Fred. 1991. The Falling Rate of Profit in the Postwar United States Economy. New York: St. Martin’s Press.

Romer, Christina D. 1999. “Changes in Business Cycles: Evidence and Explanations”. Journal of Economic

Perspectives 13(2): 23-44.

Shaikh, Anwar M. and E. Ahmet Tonak. 1994. Measuring the Wealth of Nations. Cambridge: Cambridge

University Press.

Weisskopf, Thomas E. 1979. “Marxist Crisis Theory and the Rate of Profit in the Post-War US Economy”.

Cambridge Journal of Economics 3(4): 341-78.


18

Chart 1.The Macroeconomic Rate of Profit, USA, 1948-89

0.0

0.1

0.2

0.3

0.419

48

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

Chart 2.Decomposition of the Rate of Profit: All Labor Productive

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

Rate of Surplus Value Composition of Capital


19

Chart 3.Real Profits per Hour Against the Unemployment Rate

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

0 1 2 3 4 5 6 7 8 9 10 11

Unemployment Rate (%)

$198

2

Chart 4.'Class Distributive' Rate of Surplus Value and its

Components

0.0

0.20.4

0.6

0.81.0

1.2

1.41.6

1.8

2.02.2

2.4

1948

1951

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

S/V W(u)/W(p) W(p)/W e(d)


20

Table 1

Rate of Profit-Wage VCC Rate of Profit-Wage VCCProfit Ratio Proxy Profit Ratio Proxy

1948P-1958T -3.20 -1.43 1.77 -32.05 -14.35 17.701958T-1965P 3.29 1.90 -1.39 23.06 13.33 -9.731948P-1965P -0.53 -0.06 0.47 -8.99 -1.02 7.97

1965P-1982T -4.60 -2.99 1.62 -78.28 -50.75 27.531982T-1989P 5.45 3.23 -2.22 38.15 22.63 -15.521965P-1989P -1.67 -1.17 0.50 -40.14 -28.12 12.02

1948P-1989P -1.20 -0.71 0.49 -49.13 -29.14 19.99

Decomposition of the rate of profit over profit cycles: all labor productive.Average annual change (%) Total change (%)

Table 2

Rate of Profit-Wage VCCProfit Ratio Proxy(∆%) (∆%) (∆%)

1948P-1949T -10.58 -4.17 6.411949T-1953P -0.39 -3.44 -3.061948P-1953P -10.97 -7.61 3.35

1953P-1954T -6.16 -0.92 5.241954T-1957P -12.06 -6.04 6.021953P-1957P -18.22 -6.95 11.26

1957P-1958T -2.87 0.22 3.081958T-1960P 4.19 -2.29 -6.481957P-1960P 1.32 -2.08 -3.40

1960P-1961T 2.07 2.59 0.521961T-1969P -2.47 -2.77 -0.301960P-1969P -0.40 -0.18 0.22

1969P-1970T -11.13 -7.30 3.831970T-1973P 6.00 7.54 1.541969P-1973T -5.13 0.25 5.37

1973P-1975T -20.17 -4.76 15.401975T-1979P -6.19 -9.12 -2.931973P-1979P -26.36 -13.89 12.47

1979P-1980T -13.30 -9.57 3.741980T-1981P 1.54 3.91 2.371979P-1981P -11.76 -5.65 6.11

1981P-1982T -15.77 -15.65 0.121982T-1989P 38.15 22.63 -15.521981P-1989P 22.38 6.98 -15.40

1948P-1989P -49.13 -29.14 19.99

Decomposition of the rate of profit over NBER cycles: all labor productive.


21

Table 3

r Π/W K/Wr 55.9029

Π/W 41.6419 36.1010K/W -15.2279 -6.0078 9.8731

Table 4

r Π/W K/Wr 1.0000

Π/W 0.9269 1.0000K/W -0.6482 -0.3182 1.0000

Correlation matrix for the rate of profit: all labour productive(annual rates of growth)

Variance-covariance matrix for the rate of profit: all labour productive(annual rates of growth)


22

Table 5

Rate of Rate of W(U)/W(P) S/V-W(U)/W(P) VCCProfit Surplus Value Proxy

1948P-1958T -3.20 1.78 3.44 0.07 3.281958T-1965P 3.29 1.12 0.31 2.06 -1.241948P-1965P -0.53 1.51 2.15 0.89 1.42

1965P-1982T -4.60 -0.11 1.36 -2.28 2.321982T-1989P 5.45 2.30 1.37 4.00 -1.451965P-1989P -1.67 0.59 1.36 -0.45 1.22

1948P-1989P -1.20 0.97 1.69 0.10 1.30

1948P-1958T -32.05 17.84 34.36 0.74 32.781958T-1965P 23.06 7.85 2.18 14.39 -8.671948P-1965P -8.99 25.69 36.54 15.12 24.12

1965P-1982T -78.28 -1.91 23.13 -38.81 39.471982T-1989P 38.15 16.07 9.60 27.97 -10.171965P-1989P -40.14 14.17 32.73 -10.84 29.29

1948P-1989P -49.13 39.86 69.27 4.28 53.41

Decomposition of the rate of profit over profit cycles: not all labor productive.Average annual change (%)

Total ∆%


23

Table 6

Rate of Rate of W(U)/W(P) S/V-W(U)/W(P) VCCProfit Surplus Value Proxy

1948P-1949T -10.58 2.12 6.25 -1.64 8.941949T-1953P -0.39 2.20 5.58 -1.11 -0.721948P-1953P -10.97 4.32 11.83 -2.74 8.22

1953P-1954T -6.16 5.97 8.86 2.95 9.111954T-1957P -12.06 1.30 5.57 -3.50 8.551953P-1957P -18.22 7.27 14.43 -0.55 17.66

1957P-1958T -2.87 6.25 8.09 4.03 6.901958T-1960P 4.19 0.13 1.50 -1.57 -5.761957P-1960P 1.32 6.38 9.60 2.47 1.14

1960P-1961T 2.07 2.56 1.82 3.48 1.401961T-1969P -2.47 0.34 2.02 -1.78 0.691960P-1969P -0.40 2.90 3.84 1.70 2.10

1969P-1970T -11.13 0.56 4.65 -4.97 6.161970T-1973P 6.00 -3.37 -8.40 3.38 -2.621969P-1973T -5.13 -2.80 -3.75 -1.59 3.54

1973P-1975T -20.17 5.39 9.49 -0.04 20.121975T-1979P -6.19 -4.23 -0.63 -9.44 -3.251973P-1979P -26.36 1.16 8.86 -9.49 16.87

1979P-1980T -13.30 1.73 6.61 -6.16 7.151980T-1981P 1.54 3.07 1.96 4.95 3.401979P-1981P -11.76 4.80 8.57 -1.21 10.55

1981P-1982T -15.77 -0.25 6.30 -12.27 3.501982T-1989P 38.15 16.07 9.60 27.97 -10.171981P-1989P 22.38 15.82 15.90 15.70 -6.68

1948P-1989P -49.13 39.86 69.27 4.28 53.41

Total ∆%Decomposition of the rate of profit over NBER cycles: all labor productive.


24

Table 7

r S/V W(U)/W(P) S/V-W(U)/W(P) VCC Proxyr 55.9029S/V 6.5445 6.4638W(U)/W(P) -14.1616 3.7547 10.6487S/V-W(U)/W(P) 34.3531 9.6392 -4.6909 29.7822VCC Proxy -21.7929 3.0563 9.9976 -5.9966 17.3644

Table 8

r S/V W(U)/W(P) S/V-W(U)/W(P) VCC Proxyr 1.0000S/V 0.3529 1.0000W(U)/W(P) -0.5949 0.4639 1.0000S/V-W(U)/W(P) 0.8630 0.7121 -0.2700 1.0000VCC Proxy -0.7170 0.2957 0.7536 -0.2703 1.0000

Variance-covariance matrix for the rate of profit: not all labour productive(annual rates of growth)

Correlation matrix for the rate of profit: not all labour productive(annual rates of growth)

productive and unproductive labor in the us...

Documents